🌸 Machine Learning Product Performance Analysis

Author: Ashley Prado
Dataset: product_sales.csv

---

📘 Project Overview

This project analyzes supermarket product performance using machine learning techniques. It includes:

• ✨ Full data preprocessing (missing values, outliers, normalization)
• ✨ K-means clustering fully implemented from scratch
• ✨ Regression modeling (Linear & Polynomial Regression)
• ✨ Visualizations for non-technical users
• ✨ A complete analytical write-up in REPORT.pdf

---

🛠️ Installation

Ensure Python 3.8+ is installed.

Install dependencies:

pip install pandas numpy matplotlib scikit-learn

Or with conda:

conda install pandas numpy matplotlib scikit-learn

---

▶️ How to Run the Project

1️⃣ Data Preprocessing

Handles missing values, outliers, and standardization.

python source_code/preprocessing.py

---

2️⃣ K-means Clustering (From Scratch)

Runs K-means for k = 2–8 and prints WCSS for the elbow method.

python source_code/kmeans_from_scratch.py

---

3️⃣ Cluster Analysis

Generates cluster statistics and saves results in the results/ folder.

python source_code/cluster_analysis.py

---

4️⃣ Visualizations

Creates the elbow curve and cluster scatter plot.

python source_code/visualization.py

---

5️⃣ Regression Models

Trains Linear & Polynomial Regression and saves evaluation plots.

python source_code/regression.py

---

📊 Key Findings

• 🎯 Optimal cluster count: 4
• 🔮 Best performing regression model: Polynomial Regression (degree 2)
• 📉 Model Performance:
  • Linear Regression MAE ≈ 67.97
  • Polynomial Regression MAE ≈ 1.04

---

📄 Report

A full analytical explanation of preprocessing, clustering, regression, and interpretation is provided in:

REPORT.pdf